Given increasing computing power, an important question is whether additional computational resources would be better spent reducing the horizontal grid spacing of a convection-allowing model (CAM) or adding members to form CAM ensembles. The present study investigates this question as it applies to CAM-derived next-day probabilistic severe weather forecasts created by using forecast updraft helicity as a severe weather proxy for 63 days of the 2010 and 2011 NOAA Hazardous Weather Testbed Spring Forecasting Experiments. Forecasts derived from three sets of Weather Research and Forecasting Model configurations are tested: a 1-km deterministic model, a 4-km deterministic model, and an 11-member, 4-km ensemble. Forecast quality is evaluated using relative operating characteristic (ROC) curves, attributes diagrams, and performance diagrams, and forecasts from five representative cases are analyzed to investigate their relative quality and value in a variety of situations. While no statistically significant differences exist between the 4- and 1-km deterministic forecasts in terms of area under ROC curves, the 4-km ensemble forecasts offer weakly significant improvements over the 4-km deterministic forecasts over the entire 63-day dataset. Further, the 4-km ensemble forecasts generally provide greater forecast quality relative to either of the deterministic forecasts on an individual day. Collectively, these results suggest that, for purposes of improving next-day CAM-derived probabilistic severe weather forecasts, additional computing resources may be better spent on adding members to form CAM ensembles than on reducing the horizontal grid spacing of a deterministic model below 4 km.

Spread and skill of mixed- and single-physics convection-allowing ensemble forecasts that share the same set of perturbed initial and lateral boundary conditions are investigated at a variety of spatial scales. Forecast spread is assessed for 2-m temperature, 2-m dewpoint, 500-hPa geopotential height, and hourly accumulated precipitation both before and after a bias-correction procedure is applied. Time series indicate that the mixed-physics ensemble forecasts generally have greater variance than comparable single-physics forecasts. While the differences tend to be small, they are greatest at the smallest spatial scales and when the ensembles are not calibrated for bias. Although differences between the mixed- and single-physics ensemble variances are smaller for the larger spatial scales, variance ratios suggest that the mixed-physics ensemble generates more spread relative to the single-physics ensemble at larger spatial scales. Forecast skill is evaluated for 2-m temperature, dewpoint temperature, and bias-corrected 6-h accumulated precipitation. The mixed-physics ensemble generally has lower 2-m temperature and dewpoint root-mean-square error (RMSE) compared to the single-physics ensemble. However, little difference in skill or reliability is found between the mixed- and single-physics bias-corrected precipitation forecasts. Overall, given that mixed- and single-physics ensembles have similar spread and skill, developers may prefer to implement single- as opposed to mixed-physics convection-allowing ensembles in future operational systems, while accounting for model error using stochastic methods.

Most ensembles suffer from underdispersion and systematic biases. One way to correct for these shortcomings is via machine learning (ML), which is advantageous due to its ability to identify and correct nonlinear biases. This study uses a single random forest (RF) to calibrate next-day (i.e., 12–36-h lead time) probabilistic precipitation forecasts over the contiguous United States (CONUS) from the Short-Range Ensemble Forecast System (SREF) with 16-km grid spacing and the High-Resolution Ensemble Forecast version 2 (HREFv2) with 3-km grid spacing. Random forest forecast probabilities (RFFPs) from each ensemble are compared against raw ensemble probabilities over 496 days from April 2017 to November 2018 using 16-fold cross validation. RFFPs are also compared against spatially smoothed ensemble probabilities since the raw SREF and HREFv2 probabilities are overconfident and undersample the true forecast probability density function. Probabilistic precipitation forecasts are evaluated at four precipitation thresholds ranging from 0.1 to 3 in. In general, RFFPs are found to have better forecast reliability and resolution, fewer spatial biases, and significantly greater Brier skill scores and areas under the relative operating characteristic curve compared to corresponding raw and spatially smoothed ensemble probabilities. The RFFPs perform best at the lower thresholds, which have a greater observed climatological frequency. Additionally, the RF-based postprocessing technique benefits the SREF more than the HREFv2, likely because the raw SREF forecasts contain more systematic biases than those from the raw HREFv2. It is concluded that the RFFPs provide a convenient, skillful summary of calibrated ensemble output and are computationally feasible to implement in real time. Advantages and disadvantages of ML-based postprocessing techniques are discussed.

Extracting explicit severe weather forecast guidance from convection-allowing ensembles (CAEs) is challenging since CAEs cannot directly simulate individual severe weather hazards. Currently, CAE-based severe weather probabilities must be inferred from one or more storm-related variables, which may require extensive calibration and/or contain limited information. Machine learning (ML) offers a way to obtain severe weather forecast probabilities from CAEs by relating CAE forecast variables to observed severe weather reports. This paper develops and verifies a random forest (RF)-based ML method for creating day 1 (1200–1200 UTC) severe weather hazard probabilities and categorical outlooks based on 0000 UTC Storm-Scale Ensemble of Opportunity (SSEO) forecast data and observed Storm Prediction Center (SPC) storm reports. RF forecast probabilities are compared against severe weather forecasts from calibrated SSEO 2–5-km updraft helicity (UH) forecasts and SPC convective outlooks issued at 0600 UTC. Continuous RF probabilities routinely have the highest Brier skill scores (BSSs), regardless of whether the forecasts are evaluated over the full domain or regional/seasonal subsets. Even when RF probabilities are truncated at the probability levels issued by the SPC, the RF forecasts often have BSSs better than or comparable to corresponding UH and SPC forecasts. Relative to the UH and SPC forecasts, the RF approach performs best for severe wind and hail prediction during the spring and summer (i.e., March–August). Overall, it is concluded that the RF method presented here provides skillful, reliable CAE-derived severe weather probabilities that may be useful to severe weather forecasters and decision-makers.

No abstract

No abstract

We present an overviewof recentwork on using artificial intelligence/machine learning techniques for forecasting convective weather and its associated hazards, including tornadoes, hail, wind, and lightning. These high-impact phenomena globally cause both massive property damage and loss of life yet they are quite challenging to forecast. Given the recent explosion in developing machine learning techniques across the weather spectrum and the fact that the skillful prediction of convective weather has immediate societal benefits, we present a thorough review of the current state of the art in artificial intelligence and machine learning techniques for convective hazards. Our review includes both traditional approaches, including support vector machines and decision trees as well as deep learning approaches. We highlight the challenges in developing machine learning approaches to forecast these phenomena across a variety of spatial and temporal scales. We end with a discussion of promising areas of future work for ML for convective weather, including a discussion of the need to create trustworthy AI forecasts that can be used for forecasters in real-time and the need for active cross-sector collaboration on testbeds to validate machine learning methods in operational situations.

What: Over 165 forecasters and researchers engaged in real-time, experimental severe weather forecasting activities, and evaluated convection-allowing models, including 1) several Unified Forecast System prototypes, 2) the Warn-on-Forecast System, and 3) innovative post-processing strategies.